.png){kind=logo}
Celebrating and learning from those on the frontlines of public service is core to our mission at Civic Roundtable. Our work is shaped by the voices of leaders like Miles Latham, whose efforts leading artificial intelligence for the State of Vermont begins with a focus on the public servants themselves. Our conversation below was edited for brevity.
Of course. As Vermont’s Director of Artificial Intelligence, I lead the Division of Artificial Intelligence within our state government, which in turn sits within the Agency of Digital Services, our state's centralized IT agency. Our AI Division’s role is primarily operationalizing the strategic guidance that we receive from Vermont’s Artificial Intelligence Council, which is this multidisciplinary group of leaders with diverse backgrounds who provide guidance to us from a variety of unique perspectives. We also serve as an “AI Center for Enablement” within the state by providing education and guidance on the use of AI, developing policies governing its use, and then implementing technical AI solutions across the state including tools tailored to specific agencies.
It's been a long and winding process for me. I first got into machine learning and specifically natural language processing as an undergraduate. I started working as a Python developer at a research lab at my university, then continued similar work through grad school. After, I took a position with a quant consulting firm working on data processing with really messy real-world data. We often came across patterns that would be consistent across different cases and projects, for example, in processing complicated input files like large magnitudes of PDFs or JSON files.
I was closely following the development of generative artificial intelligence from 2017 onwards and realized that there was a lot of opportunity there to use this technology to reduce the human burden for those types of repeatable data workflows.
I'm a Vermont native, and I really love this state for a multitude of reasons. So when I saw this role, I realized that it was a perfect chance to hop over to the public sector and leverage some of the expertise that I've developed to help Vermonters safely and effectively navigate this extremely salient but also deeply complex and ever-changing technology.
The current project we're working on that I'm proudest of is ChatVT. That’s our internal chatbot that we make available to state government employees on our secure internal network. I want to emphasize that our AI team did the bulk of the development work before my time — I can’t take credit for all of their fantastic work.
It's been great to have this internal chat tool that can safely be used with most types of non-public information like internal policy documents. The tool was developed internally, so it’s flexible in terms of how it can be extended to other systems. For example, we're building out a tool that helps administrators troubleshoot sign in issues for users of our state systems by automatically pulling up pertinent information for them. And there are quite a few groups throughout the state who have indicated that it's been really helpful for doing things like working with complex internal process documents.
We do strongly believe in democratizing access to AI, to the extent possible, and this ChatVT model has thus far been beneficial in that regard.
The first key opportunity that comes to mind is relatively simple and not super flashy, but that's document processing, specifically processing PDFs. Every organization that I've ever worked with has been heavily reliant on workflows involving extracting information from PDFs in some form, whether they're invoices or some other type of document. And this is one of the areas where generative AI excels, and one of the biggest opportunities because of the vast prevalence of PDFs throughout government operations, regardless of what specifically an agency or department is tasked with.
And that’s one of the main things that we want to accomplish: freeing up people to interact with our constituents and helping them solve problems, rather than spending significant time on burdensome tasks.
I've found the opportunity to meet with other government officials working in the space to be extremely helpful. It's been fascinating to see the different ways that states are approaching AI. Just in the past few months’ worth of meetings I’ve learned a great deal about how other states are looking at workforce development, for example, and piloting various AI tools that we're also interested in piloting. The lessons thus far have been extremely applicable to our work as well.
The event was extremely well organized, and I really enjoyed my time there. I think the opportunity to see the demos of all the different states’ projects was highly informative. During one exercise I mentioned to a data scientist at another state that I've been working through extracting information from these extremely long and complex PDF reports. He actually had a Python library off the top of his head that turned out to be helpful for the project. And it was great just to be able to get direct one-on-one feedback from a small group of experts from across the various states that were in attendance.
There are two things that come to mind here. The first is the importance of use case: rather than pursuing the flashiest, most cutting-edge things, it's really important to think carefully about what use cases are most salient for AI — and especially with the focus on use cases that are discrete and relatively easily to isolate.
Second, and related, is that prior to implementing AI, it’s important to fully understand the relevant existing processes. There are a lot of workflows that actually need to be modified themselves to support the use of AI before jumping into trying to build out an AI solution.
Really just having the opportunity to talk with people from diverse backgrounds and levels of experience about artificial intelligence. You know, helping to demystify everything that you hear about AI and open up that black box perception in terms of what AI actually is. In that way, as Vermonters, we can be operating from a common shared understanding of what AI is and how to use it. So just being able to sit down and talk with people and understand their perspectives, helps us bridge the information gap there — that's been really rewarding for me.
Growing up, my parents ran a horse share boarding business. So we were responsible for providing shelter and food and medical care for our customers’ horses. It wasn't strictly a farm, but it was awesome to be able to grow up around animals. We also had goats, pigs, cows, and chickens — it was a lot of work. Now my wife and I have two children, which is taking up the bulk of our bandwidth, so we’re not really doing much in regards to farming these days. I include it on my LinkedIn because although I now spend the majority of my time working with technology, farming is still a key part of who I am and how I think about the world.
Subscribe to our blog today for Roundtable news and product announcements, straight to your inbox.